U.S. patent application number 10/025797 was filed with the patent office on 2002-12-05 for adaptive rake receiving apparatus constrained with at least one constraint for use in mobile communication system and method therefor.
Invention is credited to Choi, Jinho, Hong, Seung Chul, Jung, Young-Ho, Kim, Seong Rag, Lee, Yong Hoon.
Application Number | 20020181554 10/025797 |
Document ID | / |
Family ID | 26639047 |
Filed Date | 2002-12-05 |
United States Patent
Application |
20020181554 |
Kind Code |
A1 |
Kim, Seong Rag ; et
al. |
December 5, 2002 |
Adaptive rake receiving apparatus constrained with at least one
constraint for use in mobile communication system and method
therefor
Abstract
An adaptive RAKE receiving apparatus is constrained with at
least one constraint for use in a mobile communication system. The
apparatus includes an input signal generator for generating a
complex received signal by gathering multi-path components during a
corresponding transmitting signature; adaptive filters for
filtering the complex received signal based on a tab weight that is
adjusted at a predetermined period; channel estimators for
estimating a phase component and an amplitude component of a
particular user channel by using the filtered signals to generate
channel estimating result signals; a signal recovering unit for
recovering an original signal, which was transmitted from a
particular user, by combining the filtered signals for all
multi-path components and the channel estimating result signals;
selecting unit for selecting one between a predetermined trained
data signal and the recovered signal from the signal recovering
unit; a reference signal generator for generating a reference
signal by using the selected signal and the channel estimation
result signal; an error calculator for comparing the filtered
received signal with the reference signal to calculate error
between these compared two signals; and a tap coefficient adjuster
for adjusting tap coefficients of the adaptive filtering unit based
on MMSE (Minimum Mean Square Error) criterion with at least one
constraint.
Inventors: |
Kim, Seong Rag; (Taejon,
KR) ; Choi, Jinho; (Taejon, KR) ; Jung,
Young-Ho; (Busan, KR) ; Hong, Seung Chul;
(Taejon, KR) ; Lee, Yong Hoon; (Taejon,
KR) |
Correspondence
Address: |
JACOBSON HOLMAN, PLLC.
PROFESSIONAL LIMITED LIABILITY COMPANY
400 Seventh Street, N.W.
Washington
DC
20004
US
|
Family ID: |
26639047 |
Appl. No.: |
10/025797 |
Filed: |
December 26, 2001 |
Current U.S.
Class: |
375/147 ;
375/232; 375/E1.032 |
Current CPC
Class: |
H04B 1/71055 20130101;
H04B 1/7117 20130101; H04B 1/712 20130101 |
Class at
Publication: |
375/147 ;
375/232 |
International
Class: |
H04K 001/00 |
Foreign Application Data
Date |
Code |
Application Number |
May 3, 2001 |
KR |
2001-24172 |
Jun 25, 2001 |
KR |
2001-36184 |
Claims
What is claimed is:
1. An adaptive RAKE receiving apparatus constrained with at least
one constraint in a mobile communication system, the apparatus
comprising: input signal generating means for generating a complex
received signal by gathering multi-path components during a
corresponding transmitting signature; adaptive filtering means for
filtering the complex received signal based on a tab weight that is
adjusted at a predetermined period; channel estimating means for
estimating a phase component and an amplitude component of a
particular user channel by using the filtered signal from the
adaptive filtering means to generate a channel estimating result
signal; signal recovering means for recovering an original signal,
which was transmitted from a particular user, by combining the
filtered signals from the adaptive filter means for all multi-path
components and the channel estimating result signal from the
channel estimating means; selecting means for selecting one between
a predetermined trained data signal and the recovered signal from
the signal recovering means; reference signal generating means for
generating a reference signal by using the selected signal from the
selecting means and the channel estimation result signal from the
channel estimating means; error calculating means for comparing the
filtered received signal from the adaptive filtering means with the
reference signal from the reference signal generating means to
calculate error between these compared two signals; and tap
coefficient adjusting means for adjusting tap coefficients of the
adaptive filtering means based on MMSE (Minimum Mean Square Error)
criterion with at least one constraint (constraint MMSE
criterion).
2. The receiver as recited in claim 1, wherein the constraint MMSE
criterion is defined as:J.ident.E.left
brkt-bot..vertline..sub.l(n)d.sub.-
1(n)-w.sub.l(n).sup.Hr.sub.l(n).vertline..sup.2.right brkt-bot.
subject to w.sub.l(n).sup.Hs.sub.1=1where J is the constraint MMSE
criterion, E represents a mean value, .sub.l(n) is an estimated
channel for the l-th multi-path component, d.sub.1(n) is the
selected signal from the selecting means, w.sub.l(n) is an adaptive
filter coefficient vector, and a superscript H represents Hermitian
operation, and a product of the tap coefficient w.sub.l(n) of the
adaptive filtering means and a spread code vector s, is constrained
to be substantially 1 so that the error calculated at the error
calculating means is minimized.
3. The receiver as recited in claim 2, wherein the tap coefficient
of the adaptive filtering means for the l-th multi-path component
is orthogonal-separated into a spreading code vector and an
adaptive component orthogonal to the spreading code vector
expressed as:w.sub.l(n)=s.sub.1+x.sub.l(n)where s.sub.1 is the
spreading code vector and x.sub.l(n) is the adaptive component of
the tap coefficient vector, these two vectors being orthogonal to
each other, the adaptive component orthogonal to the spreading code
vector is changed by using a component of the received signal,
which is projected into the adaptive component orthogonal to the
spreading code vector rather than using directly the received
signal.
4. The receiver as recited in claim 1, wherein the constraint MMSE
criterion is defined as:J.ident.E.left
brkt-bot..vertline.{circumflex over
(v)}.sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr(n).vertline..sup.2.right
brkt-bot. subject to w.sub.l(n)=u.sub.1l+z.sub.l(n) and
z.sub.l(n).perp.Range (U)where J is the constraint MMSE criterion,
E represents a mean value, .sub.l(n) is an estimated channel for
the l-th multi-path component, d.sub.1(n) is the output signal from
the selecting means, w.sub.l(n) is an adaptive filter coefficient
vector, and a superscript H represents Hermitian operation, the
inner product of the tap coefficient w.sub.l(n) of the adaptive
filtering means for the l-th multi-path component and the spreading
code vector s.sub.1l for the corresponding multi-path component is
constrained to be substantially 1 and the inner product of the tap
coefficient w.sub.l(n) of the adaptive filtering means for the l-th
multi-path component and the spreading code vector s.sub.1l
(l.noteq.1) for other corresponding multi-path component is
constrained to be substantially 0 so that the error that is
calculated by the error calculating means is minimized.
5. The receiver as recited in claim 4, wherein the tap coefficient
of the adaptive filter means for the l-th multi-path component is
orthogonal-separated into a spreading code vector and an adaptive
component orthogonal to the spreading code vector
as:w.sub.l(n)={overscor- e (s)}.sub.1l+x.sub.l(n)where {overscore
(s)}.sub.1l=S(S.sup.HS).sup.-1f.s- ub.l, f.sub.l is a L-by-1 column
vector with all elements 0's except 1 at the l-th position and
x.sub.l(n) is the adaptive component of the tap coefficient vector,
x.sub.l(n) being orthogonal to a range spanned by S, i.e.,
x.sub.l(n).perp.Range(S), the adaptive component orthogonal to the
spreading code vector is changed by using a component of the
received signal, which is projected into the adaptive component
orthogonal to the spreading code vector rather than using directly
the received signal.
6. The receiver as recited in claim 2, wherein the constraint MMSE
criterion for updating the coefficient of the adaptive filtering
means for the l-th multi-path component is implemented by
orthogonal separation LMS (least mean square) algorithm
as:x.sub.l(n+1)=x.sub.l(n)+.mu..multido-
t.e.sub.l(n).sup..times..multidot.P.sub.s.sup..perp.r(n)where
e.sub.l(n).ident..sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr(n), i.e.,
the difference between the product of the channel estimation value
and data and the output of the adaptive filtering means, and
P.sub.S.sup..perp.=I-S(S.sup.HS).sup.-1S.sup.H,
P.sub.S.sup..perp.r(n) being a component of r(n) projected into
x.sub.l(n), .mu. is a step size that determines the rate at which
the tap coefficient changes, and a superscript * represents complex
conjugate operation.
7. The receiver as recited in claim 1, wherein, in order to
estimate the channel for the l-th multi-path component, the channel
estimating means multiplies the outputs of the adaptive filtering
means for the multi-path components with the complex conjugate of
data for a predetermined number of the pilot symbol and averages
the multiplied results as follows: 5 c ^ i = 1 N p i = 1 N p b 1 *
( n - iQ ) w i H ( n - iQ ) r i ( n - iQ ) where N.sub.p is the
number of the pilot symbols used for the channel estimation and Q
is an inserting period of the pilot symbol.
8. The receiver as recited in claim 1, where the constraint MMSE
criterion is defined as:J.ident.E.left
brkt-bot..vertline.{circumflex over
(v)}.sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr(n).vertline..sup.2.right
brkt-bot. subject to w.sub.l(n)=u.sub.1l+z.sub.l(n) and
z.sub.l(n).perp.Range (U)where J is the constraint MMSE criterion,
E represents a mean value, {circumflex over (v)}.sub.l(n) is a
coefficient estimated by the l-th basis component, d.sub.l(n), is
the output of the selected means, z.sub.l(n) is a variable
component of an adaptive filter coefficient, r(n) is an adaptive
filter input signal vector, U=[u.sub.11 u.sub.12 . . . u.sub.1L] is
a matrix constructed by L left singular vectors of S=[s.sub.11
s.sub.12 . . . s.sub.1L] and a superscript H represents Hermitian
operation, the inner product of the tap coefficient w.sub.l(n) of
the adaptive filtering means for the l-th multi-path component and
the spreading code vector s.sub.1l for the corresponding multi-path
component is constrained to be substantially 1 and the inner
product of the tap coefficient w.sub.l(n) of the adaptive filtering
means for the l-th multi-path component and the spreading code
vector s.sub.l (l.noteq.1) for other multi-path components is
constrained to be substantially 0 so that the error that is
calculated by the error calculating means is minimized.
9. The receiver as recited in claim 8, wherein the tap coefficient
of the adaptive filtering means for the l-th multi-path component
is orthogonal-separated into a spreading code vector and an
adaptive component orthogonal to the spreading code vector
as:w.sub.l(n)=u.sub.1l+- z.sub.l(n)where u.sub.1l is the l-th left
singular vector of the spreading code matrix S and z.sub.l(n) is
the adaptive component of the tap coefficient vector, z.sub.l(n)
being orthogonal to the range spanned by U, i.e.,
z.sub.l(n).perp.Range(U), the adaptive component orthogonal to the
spreading code vector is changed by using a component of the
received signal, which is projected into the adaptive component
orthogonal to the spreading code vector rather than using directly
the received signal.
10. The receiver as recited in claim 8, wherein the constraint MMSE
criterion for updating the coefficient of the adaptive filtering
means for the l-th multi-path component is implemented by
orthogonal separation LMS (least mean algorithm) expressed
as:z.sub.l(n+1)=z.sub.l(n)+.mu..mult-
idot.e.sub.l(n)*.multidot.P.sub.U.sup..perp.r(n)where
e.sub.l(n).ident.{circumflex over
(v)}.sub.l(n)d.sub.1(n)-w.sub.l(n).sup.- Hr(n), i.e., the
difference between the product of the channel estimation and data
and the output of the adaptive filtering means,
P.sub.U.sup..perp.=I-U(U.sup.HU).sup.-1U.sup.H=I-UU.sup.H,
P.sub.U.sup..perp.r(n) is a component of r(n) projected to
z.sub.l(n), .mu. is a step size that is a rate at which the tap
coefficient is changed, and a superscript * represents complex
conjugate operation.
11. The receiver as recited in claim 1, wherein, in order to
estimate the channel for the l-th multi-path component, the channel
estimating means multiplies the outputs of the adaptive filtering
means for the multi-path components with the complex conjugate of
data for a predetermined number of the pilot symbol and averages
the multiplied values by an equation expressed as: 6 v ^ i = 1 N p
i = 1 N p b 1 * ( n - iQ ) w i H ( n - iQ ) r ( n - iQ ) where
N.sub.p is the number of the pilot symbols used for the channel
estimation and Q is an inserting period of the pilot symbol.
12. The receiver as recited in claim 1, wherein the channel
estimating means estimates the channels for all of the multi-path
components by using the outputs of the adaptive filtering means,
and the output of the selecting means for the predetermined number
of the pilot symbols expressed as: 7 [ c ^ 1 ( n ) c ^ L ( n ) ] =
[ 1 w 1 H ( n ) s 1 ( 2 - 1 ) w 1 H ( n ) s 1 ( L - 1 ) w L H ( n )
s 1 ( 1 - L ) w L H ( n ) s 1 ( 2 - L ) 1 ] - 1 [ b 1 * ( n ) w 1 H
( n ) r 1 ( n ) b 1 * ( n ) w L H ( n ) r L ( n ) ] where
s.sub.1(p) is a p chip-shifted version of s.sub.1=[s.sub.11
s.sub.12 . . . s.sub.1,N-1 s.sub.1,N].sup.T that is a normalized
spreading code for the first user, p being an arbitrary integer, if
p is a positive integer, s.sub.1(p)=[0.sub.p s.sub.11 s.sub.12 . .
. s.sub.1,N-p].sup.T and if p is a negative integer,
s.sub.1(p)=[s.sub.1,-p+1 s.sub.1,-p+2 . . . s.sub.1,N
0.sub.p].sup.T, 0.sub.p being a 1.times.p 0 vector,
(.tau..sub.i-.tau..sub.l) being the transmission delay difference
between the i-th multi-path component and the l-th multi-path
component, which is integer times of a chip.
13. An adaptive RAKE receiving method using at least one constraint
in a mobile communication system, the method comprising the steps
of: (a) setting initial coefficients of adaptive filters for
multi-path components; (b) providing each of the adaptive filters
with a corresponding multi-path component of a user to perform
complex signal filtering; (c) deciding a channel estimation value
for the multi-path component; (d) generating a reference signal by
determining a transmitted data; (e) calculating an error between
the reference signal and the filtered received signal; and (f)
updating the coefficient of the adaptive filters based on a
constraint MMSE criterion.
14. The method as recited in claim 13, wherein the step (b)
includes the steps of: (g) gathering each of the multi-path
components corresponding to the transmitted signature to provide
them to each of the adaptive filters; and (h) filtering the complex
received signal by using the input of the adaptive filter and the
coefficient of the adaptive filter.
15. The method as recited in claim 14, wherein the step (b)
includes the step of: (i) compensating transmission delays of the
multi-path components and gathering the compensated received
signals to provide them to each of the adaptive filters.
16. The method as recited in claim 14, wherein the step (g)
includes the step of: (i) gathering the received signal
corresponding to period from a starting chip of a transmitted
symbol of the firstly received multi-path component among the
multi-path components to a final chip of the transmitted symbol of
a last received multi-path component among the multi-path
components to provide the gathered signals to each of the adaptive
filters.
17. The method as recited in claim 13, wherein the step (c)
includes the steps of: (j) estimating the channel for each of the
multi-path components by using the pilot symbol; and (k)
multiplying the complex conjugate value of each of the estimated
channel value with the output of the adaptive filter for the
corresponding multi-path component, and summing up the multiplied
values for all of the multi-path components to decide channel
estimation value for the transmitted signal.
18. The method as recited in claim 17, wherein the step (j)
includes the step of: (l) estimating the channel by maximum
likelihood combination by using the outputs of the adaptive filters
for all the multi-path components and the output of the selecting
means.
19. The method as recited in claim 13, wherein the step (d)
includes the step of: (m) deciding a transmitted data; and (n)
generating a reference signal by using the decided data and the
channel value.
20. A computer readable recording medium for recording a program
for implementing in mobile communication system for providing with
an adaptive RAKE receiving apparatus constrained with at least one
constraint and having a microprocessor, the functions of: (a)
setting initial coefficients of adaptive filters for multi-path
components; (b) providing each of the adaptive filters with a
corresponding multi-path component of a user to perform complex
signal filtering; (c) determining a channel estimation value for
the multi-path component; (d) generating a reference signal by
determining a transmitted data; (e) calculating an error between
the reference signal and the filtered received signal; and (f)
updating the coefficient of the adaptive filters based on a
constraint MMSE criterion.
Description
FIELD OF THE INVENTION
[0001] The present invention is related to an adaptive RAKE
receiving apparatus constrained with at least one constraint for
use in a mobile communication system and a method therefor; and,
more particularly, to an adaptive RAKE receiving apparatus
constrained with at least one constraint for use in a mobile
communication system, for receiving user's data, each of which is
spectrally spread by using its corresponding code, and a method
therefor.
PRIOR ART OF THE INVENTION
[0002] There are disclosed conventional techniques related to a
structure of a typical adaptive MMSE (Minimum Mean Square Error)
receiver in "MMSE interference suppression for direct-sequence
spread-spectrum CDMA," U. Madow and M. L. Honig, IEEE Trans.
Commun., vol. 42, pp. 3178-3188, (December 1994) (PAPER1),
"Adaptive Detection of DS-CDMA Signals in Fading Channels," S. L.
Miller and A. N. Barbosa, IEEE Trans. Commun., vol. 46, no. 1, pp.
115-124, Jan. 1998 (PAPER2), "LMMSE Detection for DS-CDMA System in
Fading Channels," M. Latva-aho and M. Juntti, IEEE Trans. Commun.,
vol. 48, no. 2, pp. 194-199, February 2000 (PAPER3), and "A
Constrained MMSE Receiver for DS-CDMA System in Fading Channels,"
S. R. Kim, Y. G. Jeong, and I. K. Kim, IEEE Trans. Commun., vol.
48, no. 11, pp. 1793-1796, November 2000 (PAPER4).
[0003] Hereinafter, it will be described for the structures of the
receivers proposed in the above papers and their
characteristics.
[0004] A typical adaptive MMSE receiver of the PAPER1 shows good
performance when channel environment is fixed to have a simple
structure but it's performance is abruptly degraded in fading
channel environment. This phenomenon results from adaptive filters
that cannot adapt themselves for abrupt changes of phase and
amplitude of the channel. In order to solve the problem of the
typical adaptive MMSE receiver under a frequency flat fading
channel, there are proposed receivers having various structures,
which compensates for the channel changes by using a separate
channel estimation result.
[0005] The receivers of the PAPER2 to PAPER4 are some modifications
for compensating for the channel changes, which use the separate
channel estimation result. The performance of the adaptive MMSE
receiver depends on accuracy of the channel estimation value.
Typically, an adaptive filter output signal has higher SNR (Signal
to Noise Ratio) than an adaptive filter input signal. Therefore,
the adaptive filter output signal can be used in estimation of the
phase and the amplitude of the channel so as to accomplish better
performance.
[0006] In other words, in the PAPER 2, by compensating the adaptive
filter input signal for the phase change component of the channel
by estimating the phase change of the channel by using the adaptive
filter output signal, load of the adaptive filter is reduced.
However, because this scheme only compensates for the phase change
of the channel, the performance is degraded when the amplitude
change of the channel is significant.
[0007] Also, in the PAPER3, they tried to compensate for both of
the phase change and the amplitude change of the channel to improve
the performance. However, because of bias of the channel estimation
value, the adaptive filter input signal, in which multi-user
interference yet exists, rather than the adaptive filter output, is
used, which results in channel estimation performance
deterioration, which leads little improvement in bit error
rate.
[0008] Also, in the PAPER3, for a multi-path fading channel, there
are provided a linear MMSE RAKE receiver which includes MMSE
adaptive receivers, each for corresponding multi-path component,
for estimating and compensating for the phase and the amplitude of
the channel by using the adaptive filter input signal. However,
because the bias of the channel estimation value is so large that
an adaptive filter coefficient converges to zero when the channel
estimation is done by using the adaptive filter output signal, the
channel estimation should be done by using the adaptive filter
input signal. Therefore, in spite of its excellent structure, it
does not shows desired performance.
[0009] On the other hand, in the PAPER4 that has been published by
the inventors of the present invention, by using a formula for
updating a adaptive filter coefficient, there is introduced a
constrained MMSE receiver to accomplish an unbiased channel
estimation value even if the channel estimation value is obtained
by using the adaptive filter output signal in a single path fading
channel. However, since it has not addressed a multi-path fading
channel, it cannot be applied to real mobile communication
environment.
SUMMARY OF THE INVENTION
[0010] Therefore, it is an object of the present invention to
provide an adaptive RAKE receiving apparatus constrained with at
least one constraint for use in a mobile communication system, in
which phase and amplitude of each of multi-path component channels
can be estimated by using adaptive filter output signals, by
applying at least one constraint related to filter coefficients to
adaptive filter coefficient updating formula, so as to eliminate
all multi-user interference and obtain a good channel estimation
value from the adaptive filter output signals.
[0011] It is another object of the present invention to provide an
adaptive RAKE receiving method, in which phase and amplitude of
each of multi-path component channels can be estimated by using
adaptive filter output signals, by applying at least one constraint
related to filter coefficients to adaptive filter coefficient
updating formula, so as to obtain a good channel estimation
value.
[0012] It is still another object of the present invention to
provide a computer readable medium for recording a program for
implementing a function for adaptive RAKE receiving, in which phase
and amplitude of each of multi-path component channels can be
estimated by using adaptive filter output signals, by applying at
least one constraint related to filter coefficients to adaptive
filter coefficient updating formula, so as to obtain a good channel
estimation value.
[0013] In accordance with an aspect of the present invention, there
is a provided adaptive RAKE receiving apparatus constrained with at
least one constraint in a mobile communication system, the
apparatus comprising: input signal generating unit for generating a
complex received signal by gathering multi-path components during a
corresponding transmitting signature; adaptive filtering unit for
filtering the complex received signal based on a tab weight vector
that is adjusted at a predetermined period; channel estimating unit
for estimating a phase component and an amplitude component of a
particular user's channel coefficient by using the filtered signal
from the adaptive filtering unit to generate a channel estimating
result signal; signal recovering unit for recovering an original
signal, which was transmitted from a particular user, by combining
the filtered signals from the adaptive filter unit for all
multi-path components and the channel estimating result signal from
the channel estimating unit; selecting unit for selecting one
between a predetermined trained data signal and the recovered
signal from the signal recovering unit; reference signal generating
unit for generating a reference signal by using the selected signal
from the selecting unit and the channel estimation result signal
from the channel estimating unit; error calculating unit for
comparing the filtered received signal from the adaptive filtering
unit with the reference signal from the reference signal generating
unit to calculate error between these compared two signals; and tap
coefficient adjusting unit for adjusting tap coefficients of the
adaptive filtering unit based on MMSE (Minimum Mean Square Error)
criterion with at least one constraint (constraint MMSE
criterion).
[0014] In accordance with another aspect of the present invention,
there is provided a adaptive RAKE receiving method using at least
one constraint in a mobile communication system, the method
comprising the steps of: (a) setting initial coefficients of
adaptive filters for multi-path components; (b) providing each of
the adaptive filters with a desired corresponding multi-path
component of a user to perform complex signal filtering; (c)
estimating & the complex channel parameter of desired user's
each multi-path component; (d) generating a reference signal by
determining a transmitted data; (e) calculating an error between
the reference signal and the filtered received signal; and (f)
updating the coefficients of the adaptive filters based on a
constraint MMSE criterion.
[0015] In accordance with still another aspect of the present
invention, there is provided a computer readable recording medium
for recording a program for implementing in mobile communication
system for providing with an adaptive RAKE receiving apparatus
constrained with at least one constraint and having a
microprocessor, the functions of: (a) setting initial coefficients
of adaptive filters for multi-path components; (b) providing each
of the adaptive filters with a corresponding multi-path component
of a user to perform complex signal filtering; (c) determining a
channel estimation value for the multi-path component; (d)
generating a reference signal by determining a transmitted data;
(e) calculating an error between the reference signal and the
filtered received signal; and (f) updating the coefficient of the
adaptive filters based on a constraint MMSE criterion.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The above and other objects and features of the instant
invention will become apparent from the following description of
preferred embodiments taken in conjunction with the accompanying
drawings, in which:
[0017] FIG. 1 is a block diagram of the first embodiment of an
adaptive RAKE receiving apparatus constrained with at least one
constraint in a mobile communication system in accordance with the
present invention;
[0018] FIG. 2 shows a block diagram of the second embodiment of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention;
[0019] FIG. 3 is a block diagram of the third embodiment of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention;
[0020] FIG. 4 provides a diagram for showing configuration of a
receiver input signal including a number of multi-path components
in accordance with the present invention;
[0021] FIG. 5 illustrates a detailed diagram of an embodiment of a
data decision unit of a adaptive RAKE receiving apparatus
constrained with at least one constraint for use in a mobile
communication system in accordance with the present invention;
[0022] FIG. 6 offers a flow chart for an embodiment of an adaptive
RAKE receiving method using at least one constraint for use in a
mobile communication system in accordance with the present
invention; and
[0023] FIG. 7 shows a graph for bit error performance of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention.
PREFERRED EMBODIMENT OF THE INVENTION
[0024] Herein, it is assumed that a pilot channel is transmitted or
pilot symbols are periodically transmitted for channel estimation
and a spreading code for spectral spreading is a short code of
which period is equal to a spreading gain. Further, it is also
assumed that the desired user (a targeted user) is the first
user.
[0025] Further, it is also assumed that a complex valued receiving
filter output signal is a signal in unit of chip and an adaptive
RAKE receiving apparatus is constrained with at least one
constraint regarding filter coefficients. Also, it is assumed that
a multi-path component that is received earlier is processed by the
adaptive MMSE receiver having a smaller index.
[0026] FIG. 1 is a block diagram of the first embodiment of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention.
[0027] In the first embodiment, a constraint where an inner product
of a filter coefficient and a corresponding multi-path component is
maintained as 1, is applied to a coefficient updating formula for
an adaptive filter for each of multi-path components in order to
obtain a channel estimation value, that has substantially no bias,
from an adaptive filter output signal.
[0028] Referring to FIG. 1, the adaptive RAKE receiving apparatus
constrained with at least one constraint of the present invention
includes transmission delay compensation buffers 100-200,
multi-path adaptive filters 110-210, orthogonal separation LMS
(Least Mean Square) filter coefficient updaters 120-220, multi-path
channel estimators 130-230, a multi-path component combiner 340, a
data decision unit 370 and a data selector 380.
[0029] The adaptive RAKE receiving apparatus further includes a
number of channel estimators 160-260, multipliers and adders.
[0030] At first, an output signal r(m) of a receiving filter (not
shown), having multi-path components of the desired user, the
multi-path components having multi-path transmission delay
different from each other, passes through the first to the L-th
transmission delay compensation buffers 100-200, each of which
compensates the corresponding multi-path transmission delay and
provides an N-chip output signal of the receiving filter
corresponding to the receiving duration of the corresponding
symbol.
[0031] Herein, assuming that there are two multi-path components
and a relative transmission delay of a lately received (the second)
multi-path component is 5 chips relative to the other received (the
first) multi-path component, the transmission delay compensation
buffers compensate the transmission delays of all multi-path
components for a maximum transmission delay.
[0032] That is, the transmission delay compensation buffer 100 for
the first multi-path component delays the receiving filter output
signal by time corresponding to 5 chips and then provides the first
adaptive filter 110 with successive N chips of the delayed
receiving filter output signal r.sub.1(n)=[r(m-5) r(m-5) . . .
r(m-5+N-1)].sup.T. In this, N.times.1 vector r.sub.1(n) includes
the received signal of N chips corresponding to a duration in which
the first multi-path component of the desired user's n-th
transmitted data symbol is received.
[0033] Similarly, the transmission delay compensation buffer 200
for the second multi-path component provides the second adaptive
filter 120 with successive N chips of the non-delayed receiving
filter output signal r.sub.2(n)=[r(m) r(m) . . . r(m+N-1)].sup.T.
In this, N.times.1 vector r.sub.2(n) includes the received signal
of N chips corresponding to the duration in which the second
multi-path component of the desired user's n-th transmitted data
symbol is received.
[0034] Herein, the output of the first transmission delay
compensation buffer 100 for the first multi-path component is
multiplied by a tap coefficient w.sub.1(n) at the first adaptive
filter 110. Similarly, the output of the L-th transmission delay
compensation buffer 200 for the L-th multi-path component is
multiplied by a tap coefficient w.sub.L(n) at the multi-path
adaptive filter 210. The tap size of each of the adaptive filters
110-210 is N taps that is equal to the size of the filter input
signal vector.
[0035] Then, channel estimation is performed for each of the
multi-path channels. The channel estimation value is obtained by
averaging, during a predetermined period, temporary channel
estimation values that are obtained by eliminating data component
from the output of each of the multi-path adaptive filters 110-210,
respectively, by using pilot symbols.
[0036] That is, the average value of the outputs of the 1-th
adaptive filter for the n-th transmitted symbol is
c.sub.1(n)b.sub.1(n)+inter-path interference. Since the inter-path
interference is relatively small, the temporary channel estimation
value can be obtained by multiplying the complex conjugate of the
transmitted symbol, b.sub.1*(n), that is known from the pilot
symbol.
[0037] Then, to reduce the effect of noise, the final channel
estimation is obtained by averaging the temporal channel estimation
values that are obtained by a number of pilot symbols. At this
time, the final channel estimation value is as follows: 1 c ^ i = 1
N p i = 1 N p b 1 * ( n - iQ ) w i H ( n - iQ ) r i ( n - iQ ) Eq .
1
[0038] where N.sub.p is the number of successive pilot symbols used
for the channel estimation and Q is a inserting period of the pilot
symbol.
[0039] The channel estimator 130 for the first multi-path component
separately estimates the channel estimation value for the first
multi-path component depending on the above estimation scheme (by
using Eq. 1) from the output of the adaptive filter 110 for the
first multi-path component and the output of the data selector
380.
[0040] Similarly, the channel estimator 230 for the L-th multi-path
component estimates the channel parameter for the L-th multi-path
component depending on the above estimation scheme in Eq. 1 from
the output of the adaptive filter 210 for the L-th multi-path
component and the output of the data selector 380.
[0041] On the other hand, the filter coefficient of each of the
adaptive filters 110-210 are updated at the transmission symbol
rate by using the orthogonal separation LMS filter coefficient
updaters 120-220 having a number of constraints.
[0042] Typically, as an algorithm for changing tap coefficients of
the adaptive filters, the LMS algorithm can be used and the tap
coefficient of the adaptive filter for the l-th multi-path
component is updated by using the conventional LMS algorithm as
follows:
w.sub.1(n+1)=w.sub.1(n)+.mu.[.sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr.sub.l(n-
)]*r.sub.l(n) Eq. 2
[0043] where .mu. the step size at which the tap coefficient of the
adaptive filter is changed, .sub.l(n) is the channel estimation
value of the l-th multi-path component, and d.sub.l(n) is the
output value of the data selector 380 for the n-th transmitted
signal of the first user.
[0044] Therefore, with adapting the filter coefficient by using
this, the filter coefficient is adapted so as to minimize the mean
square error as follows:
J.ident.E.left
brkt-bot..vertline..sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr.su-
b.l(n).vertline..sup.2.right brkt-bot. Eq. 3
[0045] However, when using the conventional LMS algorithm that
estimates the channel by using the adaptive filter output signal as
in the receiver of the present invention, the tap coefficients of
the multi-path adaptive filters 110-210 converge to zero as in a
single path fading channel as in the PAPER4.
[0046] To use the conventional LMS algorithm, the input signals to
the multi-path channel estimators 130-230 in the structure proposed
in the PAPER3 are changed to be the input signals of the multi-path
adaptive filters 110-210. However, this structure change degrades
the performance of the receiving apparatus.
[0047] In the present invention, to solve the above problem of the
conventional LMS algorithm, a constraint is applied to the filter
coefficient updating formula. This can be referred to an extension
of the method of the PAPER4.
J.ident.E.left
brkt-bot..vertline..sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr.su-
b.l(n).vertline..sup.2.right brkt-bot. subject to
w.sub.l(n).sup.Hs.sub.1=- 1 Eq. 4
[0048] where s.sub.1 is a normalized spreading code vector and a
superscript H represents Hermitian operation.
[0049] To implement efficiently the constraint as in Eq. 4, and the
tap coefficient w.sub.l(n) of the adaptive filter for the l-th
multi-path component, the orthogonal separation scheme is used as
follows:
w.sub.l(n)=s.sub.1+x.sub.l(n) Eq. 5
[0050] That is, the inner product of the tap coefficient w.sub.l(n)
of the adaptive filter and the spreading code s.sub.l always has 1
i.e., w.sub.l(n).sup.Hs.sub.1=1, as follows.
Here,.parallel.s.parallel..sup.2 is 1 since s.sub.1 is the
normalized spreading code.
w.sub.l(n).sup.Hs.sub.1=(s.sub.1+x.sub.l(n)).sup.Hs.sub.1=s.sub.1.sup.Hs.s-
ub.1=.parallel.s.sub.1.parallel..sup.2=1 Eq. 6
[0051] From this result, the final filter coefficient updating
formula can be expressed by an equation (7) as follows:
x.sub.l(n+1)=x.sub.l(n)+.mu..multidot.e.sub.l(n)*.multidot.r.sub.lx(n)
Eq. 7
[0052] where
e.sub.l(n).ident..sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr.sub.1(- n)
is difference between a product of the channel estimation value and
data and the output of the adaptive filter, and
r.sub.lx(n)=r.sub.l(n)-s.- sub.1.sup.Hr.sub.l(n)s.sub.1 is the
component of the received signal r.sub.l(n) projected to the
adaptation component x.sub.l(n) of the tap coefficient.
[0053] On the other hand, the orthogonal separation LMS filter
coefficient updaters 120-220 are provided with the adaptive filter
input signal r.sub.l(n) and e.sub.l(n), which is used to update the
filter coefficient by using Eq. 7.
[0054] The error signal e.sub.1(n) for the first adaptive filter
110 is generated by multiplying the output of the data selector 380
with the channel estimation value .sub.l(n) for the first
multi-path component at the multiplier 150 and subtracting the
output of the first multi-path adaptive filter 110 from the
multiplied values.
[0055] The error signal e.sub.L(n) for the L-th adaptive filter 210
is similarly generated by multiplying and subtracting signals.
[0056] On the other hand, diversity gain is accomplished by
multiplying the output of the multi-path adaptive filter with a
weighting value and then summing up the multiplied values. In the
present invention, the complex conjugate value of the channel
estimation value for each multi-path component is calculated at the
corresponding one of the complex conjugate calculators 160-260 and
this complex conjugate value is multiplied with the output of the
corresponding one of the multi-path adaptive filters 110-210 as a
weighing value. The outputs of the multipliers 151-251, i.e., all
of the multi-path components, are combined at the multi-path
component combiner 340. At this time, Maximum Ratio Combining
scheme is employed.
[0057] After the energy of all multi-path components is combined at
the multi-path combiner 340, transmitted data symbol is decided
from the combined output at the data decision unit 370. The data
decision unit 370 has its structure as shown in FIG. 5.
[0058] FIG. 5 illustrates a detailed diagram of an embodiment of a
data decision unit of an adaptive RAKE receiving apparatus
constrained with at least one constraint for use in a mobile
communication system in accordance with the present invention.
[0059] As shown in FIG. 5, because the input signal of the data
decision unit 340 is a complex signal, this complex signal is
divided into its real part and its imaginary part at a real
calculator 371 and an imaginary calculator 372, respectively. Then,
distances from the complex signal to all of signatures on
modulation constellation are obtained at a distance calculator 373.
From these distances, a symbol selector 374 decides a symbol that
has minimum error and outputs a complex value corresponding to the
decided signature.
[0060] When BPSK (Binary Phase Shift Keying) is used, because there
is data only at the real part and the signature is 1 or -1, the
above procedure can be simplified to only taking the real part of
the input signal of the data decision unit and determining the
signature of the real part.
[0061] Typically, in order that the adaptive receiver converges to
an optimum solution, the filter coefficient is adapted by training
procedure. Because the transmitted signature is known during the
training period or pilot transmission period, the data selector 380
selects data b.sub.1(n) regardless of decision of the data decision
unit 370. During remaining periods, the data selector 380 selects a
decided bit {circumflex over (b)}.sub.1(n). Using the pilot symbol
that is constantly inserted, as training data, in the present
invention, there is no need for additional training data that is
unless required for the typical adaptive receiver.
[0062] FIG. 2 shows a block diagram of the second embodiment of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention.
[0063] As shown in FIG. 2, the adaptive RAKE receiving apparatus
constrained with the constraint of the present invention includes
the first to the L-th transmission delay compensation buffers
100-200, the first to the L-th multi-path adaptive filters 110-210,
the first to the L-th orthogonal separation LMS filter coefficient
updaters 120-220, a maximum likelihood channel estimator 330, a
multi-path component combiner 340 and a data selector 380.
[0064] The adaptive RAKE receiving apparatus further comprises a
number of channel estimators 160-260, multipliers and adders.
[0065] In this case, the channel estimation for all of the
multi-path components are simultaneously performed at the maximum
likelihood estimator 330. The adaptive filter outputs for all of
the multi-path components and the output of the data decision unit
380 are inputted to the maximum likelihood channel estimator
330.
[0066] Herein, it will be described for a principle of the
operation of the maximum likelihood channel estimator 330.
[0067] In other words, assuming that the sum of the multi-user
interference and noise has Gaussian distribution, the value that is
resulted from multiplication the output of each of the L adaptive
filters with the conjugate complex of data has Gaussian
distribution with a mean as follows: 2 E [ b 1 * ( n ) w i H ( n )
r i ( n ) ] = i = 1 L c i ( n ) w i H ( n ) s 1 ( i - i ) Eq .
8
[0068] where s.sub.1(p) is a p-chip-shifted version of a normalized
spreading code s.sub.1=[s.sub.1,1 s.sub.1,2 . . . s.sub.1,N-1
s.sub.1,N].sup.T for the first user, p being an arbitrary integer.
When p is a positive integer, s.sub.1(p) can be expressed as
follows:
s.sub.1(p)=[0.sub.p s.sub.11 s.sub.12 . . . s.sub.1,N-p].sup.T Eq.
9
[0069] When p is a negative integer, s.sub.1(P) can be expressed as
follows:
s.sub.1(p)=[s.sub.1,-p+1 s.sub.1,-p+2 . . . s.sub.1,N
0.sub.p].sup.T Eq. 10
[0070] where 0.sub.p is a 1.times.p 0 vector.
(.tau..sub.i-.tau..sub.l) is a transmission delay difference
between the i-th multi-path component and the l-th multi-path
component and it is assumed to be integer times of a chip. For all
of the multi-path components, the maximum likelihood channel
estimation value can be obtained, if using the L terms as expressed
in Eq. 8, as follows: 3 [ c ^ 1 ( n ) c ^ L ( n ) ] = [ 1 w 1 H ( n
) s 1 ( 2 - 1 ) w 1 H ( n ) s 1 ( L - 1 ) w L H ( n ) s 1 ( 1 - L )
w L H ( n ) s 1 ( 2 - L ) 1 ] - 1 [ b 1 * ( n ) w 1 H ( n ) r 1 ( n
) b 1 * ( n ) w L H ( n ) r L ( n ) ] Eq . 11
[0071] By using a predetermined number of the pilot symbols, the
obtained values are averaged. By using this, though unbiased
channel estimation values can be obtained simultaneously for all of
the multi-path components, calculation complexity increases due to
calculation of an inverse matrix, compared with the first
embodiment.
[0072] FIG. 3 is a block diagram of the third embodiment of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention.
[0073] In the third embodiment, with maintaining the inner product
of the filter coefficient and the corresponding multi-path
component as 1 in the adaptive filter coefficient updating formula,
other constraints such as maintaining the inner product of the
filter coefficient and the inter-path interference as 0 are
employed to eliminate all of user interference so as to obtain the
unbiased channel estimation value from the adaptive filter
output.
[0074] As shown in FIG. 3, the adaptive RAKE receiving apparatus
constrained with multiple constraints of the present invention
comprises an input signal generating buffer 101, the first to the
L-th multi-path adaptive filters 110-210, a plurality of
constraints orthogonal separation LMS filter coefficient updaters
120-220, the first to the L-th multi-path channel estimators
130-230, a multi-path component combiner 340, a data decision unit
370, and a data selector 380.
[0075] The adaptive RAKE receiving apparatus further comprises a
number of channel estimators 160-260, multipliers, and adders.
[0076] The receiving filter output signal r(m) is applied to the
input signal generating buffer 101 in order to generate an adaptive
filter input signal vector having an adequate form. Offset values
for the multi-path components are provided from a searcher that
searches the offset values of effective multi-path components in a
typical DS-CDMA system.
[0077] The input signal generating buffer 101 gathers N+M chips of
the received signal from a starting chip of a n-th transmission
symbol of the firstly received multi-path component to the final
chip of the n-th transmission symbol of the lastly received
multi-path component to provide them to each of the adaptive
filters.
[0078] In other words, taking the firstly received component as a
reference, when the output of the received signal corresponding to
the starting chip of the n-th transmission symbol is r(nN+1) and
maximum difference among the transmission delays is M chips, the
output signals of the input signal generating buffer 101, which is
a (N+M).times.1 vector, for the n-th transmission symbol and the
n+1-th transmission symbol are expressed, respectively, as
follows:
r(n)=[r(nN+1) r(nN+2) . . . r(nN+N+M)].sup.T Eq. 12
r(n+1)=[r((n+1)N+1) r((n+1)N+2) . . . r((n+2)N+M)].sup.T Eq. 13
[0079] The output signal r(n) of the input signal generating buffer
101 for the n-th transmitted signal is inputted to each of the
multi-path adaptive filters 110-210, in which r(n) is multiplied
with the adaptive filter coefficient. When a (N+M).times.1
coefficient vector of the l-th adaptive filter is referred to
w.sub.l(n), the output of the adaptive filter 110 for the first
multi-path component is equal to w.sub.l.sup.H(n)r(n), the output
of the adaptive filter for the l-th multi-path component is equal
to w.sub.l.sup.H(n)r(n), and the output of the adaptive filter 210
for the L-th multi-path component is equal to
W.sub.L.sup.H(n)r(n).
[0080] Then, the channel is estimated by averaging, for a
predetermined period, the temporal channel estimation values that
are obtained by eliminating data component from the outputs of the
multi-path adaptive filters 110-210 by using the pilot symbol.
[0081] That is, because the average value of the l-th multi-path
adaptive filter for the n-th transmitted signature is
c.sub.lb.sub.1(n), the temporal channel estimation values can be
obtained by multiplication with the complex conjugate b.sub.1*(n)
of the transmitted signature that is known from the pilot
symbol.
[0082] Then, the final channel estimation value is obtained by
averaging the temporal channel estimation values obtained by using
a number of the pilot symbols, by using the method in Eq. 1.
[0083] The output of the adaptive filter 110 for the first
multi-path component and the output of the data selector 380 are
applied to the channel estimator 130 for the first multi-path
component where the channel estimation value (Eq. 1) for the first
multi-path component of the first user is estimated by using those
two outputs based on the prescribed estimation method.
[0084] Then, similarly, the output of the adaptive filter 210 for
the L-th multi-path component and the output of the data selector
380 are applied to the channel estimator 230 for the L-th
multi-path component where the channel estimation value is
separately estimated by using Eq. 1.
[0085] On the other hand, the filter coefficients of the adaptive
filters 110-210 are updated at a transmitted signature rate by
using the plural-constraints orthogonal separation LMS filter
coefficient updaters 120-220.
[0086] The algorithm for the plural-constraints orthogonal
separation LMS filter coefficient updaters can be implemented in
two ways.
[0087] Firstly, several parameters and operators are defined before
description of the algorithm.
S=[s.sub.11 s.sub.12 . . . s.sub.1L] Eq. 14
P.sub.A.sup..perp.=I-A(A.sup.HA).sup.-1A.sup.H Eq. 15
[0088] Eq. 14 is an N.times.L matrix including the spreading codes
of the targeted users for the multi-path components. Each of the
column vectors is defined as the same as in Eq. 7. Eq. 15 is an
N.times.N orthogonal complementary projector. When any N.times.L
matrix is multiplied with Eq. 15, the component of the multiplied
matrix, orthogonal to all column vectors of the A matrix is
obtained.
[0089] In the first structure for implementing the algorithm, the
filter coefficient of the adaptive filter 110 for the l-th
multi-path component converges so as to minimize following Eq.
16.
J.ident.E.left
brkt-bot..vertline..sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr(n)-
.vertline..sup.2.right brkt-bot. subject to w.sub.l(n)={overscore
(s)}.sub.1l+x.sub.l(n) and x.sub.l(n).perp.Range(S), where
{overscore (s)}.sub.1l=S(S.sup.HS).sup.-1f.sub.l. Eq. 16
[0090] In Eq. 16, f.sub.l is a L-by-1 column vector with all
elements 0's except 1 at the l-th position. In considering the
constraints, because x.sub.l(n) is orthogonal to the spreading code
for the l-th multi-path component as well as other multi-path
components, the bias of the channel estimation due to the
multi-path interference is eliminated. Adaptation algorithm to
minimize Eq. 16 is obtained by following Eq. 17.
x.sub.l(n+1)=x.sub.l(n)+.mu..multidot.e.sub.l(n)*.multidot.P.sub.s.sup..pe-
rp.r(n) Eq. 17
[0091] where
e.sub.l(n).ident..sub.l(n)d.sub.1(n)-w.sub.l(n).sup.Hr(n), i.e.,
the difference between the product of the channel estimation value
and data and the output of the adaptive filter, and
P.sub.S.sup..perp.=I-S(S.sup.HS).sup.-1S.sup.H. In case that the
channel estimation is performed in Eq. 1, the unbiased channel
estimation value can be obtained so that performance is improved. 4
E [ c ^ i ] = E [ 1 N p i = 1 N p b 1 * ( n - iQ ) w i H ( n - iQ )
r ( n - iQ ) ] = E [ 1 N p i = 1 N p b 1 * ( n - iQ ) ( s _ 1 i + x
i ( n - iQ ) ) H r ( n - iQ ) ] = E [ 1 N p i = 1 N p ( s _ 1 i + x
i ( n - iQ ) ) H i = 1 L c i s 1 i ] = c i Eq . 18
[0092] By adapting the filters based on the plural-constraints
orthogonal separation LMS filter coefficient updating algorithm,
i.e., multiplying the filter output with the complex conjugate of
the estimation value of a corresponding channel component, the
product of the energy of the correspond multi-path component and
data is generated. At the data combiner 340, this product is added
to other products of the energy of other multi-path components and
data to obtain the estimation value for the transmitted data of the
targeted user.
[0093] A vector of data component in a Range(S) spanned by L
spreading code vectors (a sum of all of the multi-path components),
each code vector corresponding to each multi-path component, can be
considered as a linear sum of each spreading code vector as a
basis. The series of receiving procedures can be considered as a
procedure for obtaining weighting values for each basis by using
the multi-path adaptive filters.
[0094] If the signal component in a same Range is expressed by
using the L bases rather than by the spreading code, a new plural
constraints orthogonal separation LMS filter coefficient updating
algorithm that is completely equivalent to the former algorithm can
be derived.
[0095] By constructing an N.times.L matrix U=[u.sub.11 u.sub.12 . .
. u.sub.1L] with L left singular vectors of S by singular value
decomposition of the code matrix S, the range spanned by U is
completely identical to the range spanned by S and column vectors
are orthogonal to each other. Therefore, the signal of the targeted
user that is a sum of multi-path components can be expressed as a
linear summation of column vectors of U as bases.
[0096] In this case, if Eq. 19 is satisfied and the channel is
estimated by the method of Eq. 1 by using the filter output,
v.sub.l(n) is estimated rather than the multi-path component.
Therefore, the outputs of the channel estimators 130-230 are
expressed as follows:
Sc(n)=Uv(n) Eq. 19
[0097] where v(n)=U.sup.HSc(n).
[0098] Based on this, Eq. 16 can be expressed as follows:
J.ident.E.left brkt-bot..vertline.{circumflex over
(v)}.sub.1(n)d.sub.1(n)-
-w.sub.l(n).sup.Hr(n).vertline..sup.2.right brkt-bot. subject to
w.sub.l(n)=u.sub.1l+z.sub.l(n) and z.sub.l(n).perp.Range (U) Eq.
20
[0099] Further, using this, the filter updating formula is obtained
as follows:
z.sub.l(n+1)=z.sub.l(n)+.mu..multidot.e.sub.l(n)*.multidot.P.sub.U.sup..pe-
rp.r(n) Eq. 21
[0100] where e.sub.l(n).ident.{circumflex over
(v)}.sub.l(n)d.sub.1(n)-w.s- ub.l(n).sup.Hr(n), i.e., the
difference between the product of the channel estimation value and
data and the output of the adaptive filter, and
P.sub.U.sup..perp.=I-U(U.sup.HU).sup.-1U.sup.H=I-UU.sup.H.
[0101] As described above, the adaptive filter input signal r(n)
and error signal e.sub.l(n) are applied to the plural-constraints
orthogonal separation LMS filter coefficient updating algorithm
blocks 120-220 where the filter coefficient is updated by using Eq.
17 or Eq. 21.
[0102] The error signal e.sub.l(n) for the first adaptive filter is
generated by multiplying 150 the output of the data selector 380
with the channel estimation value for the first multi-path
component .sub.1(n) (in Eq. 20, {circumflex over (v)}.sub.1(n) )and
subtracting the output signal of the adaptive filter 110 for the
first multi-path component from the multiplied value.
[0103] Similarly, the error signal e.sub.l(n) of the L-th adaptive
filter is generated by the multiplier 250 and the adder 240.
[0104] On the other hand, diversity benefit can be obtained from
multiplying an adequate weighting value with the output signal of
the adaptive filter for each of the multi-path components and
adding the weighted output signal. In the present invention, after
the complex conjugates of the channel estimation values for
multi-path components are obtained at the complex conjugate
calculators 160-260, the complex conjugates are multiplied with the
outputs of the multi-path adaptive filters 110-210 by the
multipliers 151-251, respectively. The outputs of the multipliers
151-251, i.e., all of the multi-path components, are summed up by
the multi-path component combiner 340 by employing maximum ratio
combination scheme.
[0105] Then, the energy of all of the multi-path components summed
up at the multi-path component combiner 340 and then transmitted
data symbol is temporally decided by the data decision unit
370.
[0106] FIG. 4 provides a diagram for showing configuration of a
receiver input signal including a number of multi-path components
in accordance with the present invention.
[0107] FIG. 6 offers a flow chart for an embodiment of an adaptive
RAKE receiving method using one constraint for use in a mobile
communication system in accordance with the present invention.
[0108] As shown in FIG. 6, the adaptive RAKE receiving apparatus
constrained with at least constraints of the present invention sets
601 the initial coefficient of each of the adaptive filters for the
L multi-path components as the spreading code of the corresponding
multi-path component and initializes the index n to 1.
[0109] The spreading code is s.sub.1 in the first and the second
embodiments and S.sub.1l in the third embodiment.
[0110] The adaptive RAKE receiving apparatus of the present
invention transfers 605 the corresponding multi-path component of
the user to the adaptive filter. In the first and the second
embodiments, at the transmission delay buffers 100-200,
transmission delays of the multi-path components are compensated to
gather the received signals during the transmitted symbol period
and apply the gathered signals to the adaptive filter. In the third
embodiment, at the input signal generating buffer 101, the received
signals corresponding to the starting chip of the multi-path
component that is firstly received among the multi-path components
to the final chip of the multi-path component that is lastly
received are gathered 607 to apply to the adaptive filters,
respectively.
[0111] The channel of each multi-path is estimated in 609 by
averaging, during a predetermined period, the temporal channel
estimation values which are obtained by eliminating the data
component from the outputs of the adaptive filters 110-210 for the
multi-path components, respectively, by using the pilot symbol. In
the first and the third embodiments, the channels are estimated at
the multi-path channel estimators. In the second embodiment, the
multi-path channels are estimated at the maximum likelihood channel
estimators.
[0112] The soft decision value for the transmitted signal is
obtained in 611 by multiplying the complex conjugate of the channel
estimation values with the outputs of the adaptive filter,
respectively, and summing up the multiplied values for all of the
multi-path components.
[0113] Then, after the n-th transmitted data is decided in 613, the
reference signal is generated 615 by using the decided data and the
channel estimation value.
[0114] Then, the difference between the signals is produced 617 by
comparing the reference signal with the filtered received signal
and the coefficients of the multi-path adaptive filters are updated
619 based on the MMSE criterion with at least constraints.
[0115] Upon completing data reception, the adaptive RAKE receiving
method of the present invention is ended, whereas the initialized
index is increased 621 to repeat the above procedure while data
reception continues.
[0116] FIG. 7 shows a graph for bit error performance of an
adaptive RAKE receiving apparatus constrained with at least one
constraint for use in a mobile communication system in accordance
with the present invention.
[0117] That is, FIG. 7 illustrates a simulation result for
explaining good data detection performance of the present
invention. Herein, for the simulation environment, it is assumed
that the channel is a Rayleigh Fading channel having 3 multi-paths,
the number of the users having same transmission power is 5, Gold
code having its period of 31 is used as the spreading code, and
fading change rate expressed by a product of Doppler frequency
f.sub.D and signature period T, i.e., f.sub.DT is a value from
10.sup.-4 to 10.sup.-2.
[0118] In FIG. 7, a reference character A represents the
conventional RAKE receiver, a reference character B represents the
adaptive MMSE receiver as in PAPER3, a reference character C
represents the first embodiment of the present invention, and a
reference character D represents the second and third embodiment of
the present invention.
[0119] As described above, the method of the present invention may
be implemented in a program to be stored at a computer readable
recording medium such as CD ROMs, RAMs, ROMs, floppy disks, hard
disks and magneto-optical disks.
[0120] As described above, the present invention may be applied to
a CDMA receiver so that data detection performance and quality of
service are improved and high speed operation is available. Also,
service area of a base station is enlarged in the multi-path fading
channel environment, than the conventional RAKE receiver or other
adaptive MMSE receivers which have been developed for the
multi-path fading channel.
[0121] Further, since the multi-user interference is eliminated in
the present invention, there is no need for strict power
control.
[0122] On the other hand, the present invention requires only
information for the targeted user (the spreading code,
synchronization information) whereas information for all users are
required in the conventional interference eliminator. Therefore,
the present invention may be also applied to a mobile terminal as
well as the base station.
[0123] While the present invention has been shown and described
with respect to the particular embodiments, it will be apparent to
those skilled in the art that many changes and modifications may be
made without departing from the spirit and scope of the invention
as defined in the appended claims.
* * * * *